This repository was archived by the owner on Jul 1, 2023. It is now read-only.
[Layers] Update DepthwiseConv2D(filterShape:...) bias initialization
#441
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thanks for the patch. Could you please add a test? |
rxwei
approved these changes
Aug 31, 2019
rxwei
reviewed
Aug 31, 2019
ocampor
pushed a commit
to ocampor/swift-apis
that referenced
this pull request
Oct 4, 2019
…ion (tensorflow#441) **Issue**: `DepthwiseConv2D(filterShape:strides:padding:activation:filterInitializer:...)` produces a shape mismatch error in the conv-bias sum of `callAsFunction()` when initialized under a `filterShape` with channel-multiplier greater than `1`. More specifically, this is caused by initializing the `bias` vector with dimension equal to said channel-multiplier. **Fix** :This PR adds the proper `bias` vector initialization by computing the dimensionality as: `bias vector dimension = output channel count = (input channel count * channel multiplier)`
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Issue:
DepthwiseConv2D(filterShape:strides:padding:activation:filterInitializer:...)produces a shape mismatch error in the conv-bias sum of
callAsFunction()when initialized under afilterShapewith channel-multiplier greater than1.More specifically, this is caused by initializing the
biasvector with dimension equal to said channel-multiplier.Fix :This PR adds the proper
biasvector initialization by computing the dimensionality as:bias vector dimension = output channel count = (input channel count * channel multiplier)